Non-Cochrane · intervention · PM42121169
Effectiveness of artificial intelligence-driven simulation in dental education: a systematic review and meta-analysis of learning outcomes.
Gadbail Amol Ramchandra, Gondivkar Shailesh M, Yuwanati Monal B, et al. · BMC medical education · 2026 · PubMed · doi:10.1186/s12909-026-09390-6
Findings
22 findings
| Population | ||||
|---|---|---|---|---|
| Dental students (undergraduate and postgraduate) | AI-driven simulation (all modalities) | Traditional training (phantom-head, lecture, video, non-AI simulation) | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Sensitivity analysis excluding studies with imputed SDs (Sherif 2026, Vannaprathip 2025) | SMD 0.94 [0.42, 1.46] GRADE not reported 3 studies |
| Dental students (undergraduate and postgraduate) | AI-driven simulation (all modalities) | Traditional training (phantom-head, lecture, video, non-AI simulation) | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Leave-one-out sensitivity analysis: omit Sherif 2026 | SMD 1.12 [0.70, 1.55] GRADE not reported Studies not reported |
| Dental students (undergraduate and postgraduate) | AI-driven simulation (all modalities) | Traditional training (phantom-head, lecture, video, non-AI simulation) | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Leave-one-out sensitivity analysis: omit Usta_Mobile 2026 | SMD 1.24 [0.83, 1.65] GRADE not reported Studies not reported |
| Dental students (undergraduate and postgraduate) | AI-driven simulation (all modalities) | Traditional training (phantom-head, lecture, video, non-AI simulation) | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Leave-one-out sensitivity analysis: all omissions combined (random effects HK) | SMD 1.22 [1.11, 1.33] GRADE not reported Studies not reported |
| Dental students (undergraduate and postgraduate) | AI-driven simulation (all modalities) | Traditional training (phantom-head, lecture, video, non-AI simulation) | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Leave-one-out sensitivity analysis: omit Huang 2025 | SMD 1.34 [1.07, 1.62] GRADE not reported Studies not reported |
| Dental students (undergraduate and postgraduate) | AI-driven simulation (all modalities) | Traditional training (phantom-head, lecture, video, non-AI simulation) | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Overall pooled estimate | SMD 1.20 [0.73, 1.67] GRADE not reported 5 studies · 557 participants |
| Dental students (undergraduate and postgraduate) | AI-driven simulation (all modalities) | Traditional training (phantom-head, lecture, video, non-AI simulation) | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Leave-one-out sensitivity analysis: omit Usta_VR 2026 | SMD 1.17 [0.77, 1.57] GRADE not reported Studies not reported |
| Dental students (undergraduate and postgraduate) | AI-driven simulation (all modalities) | Traditional training (phantom-head, lecture, video, non-AI simulation) | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Leave-one-out sensitivity analysis: omit Vannaprathip 2025 | SMD 1.09 [0.76, 1.43] GRADE not reported Studies not reported |
| Dental students (undergraduate and postgraduate) | AI-driven simulation (all modalities) | Traditional training (phantom-head, lecture, video, non-AI simulation) | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Leave-one-out sensitivity analysis: omit Zhang 2026 | SMD 1.24 [0.79, 1.70] GRADE not reported Studies not reported |
| Dental students (undergraduate and postgraduate) | AI-driven simulation (all modalities) | Traditional training (phantom-head, lecture, video, non-AI simulation) | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Trim-and-fill adjusted (2 imputed studies) | SMD 1.04 [0.57, 1.51] GRADE not reported 5 studies |
| Dental students (undergraduate and postgraduate) | VR-based AI simulation (haptic and adaptive VR) | Traditional training | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Subgroup by AI modality: VR | SMD 1.49 [0.99, 2.00] GRADE not reported 2 studies · 141 participants |
| Dental students (undergraduate and postgraduate) | Mobile AI application | Traditional training | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Subgroup by AI modality: Mobile AI | SMD 0.88 [-0.25, 2.01] GRADE not reported 1 study · 15 participants |
| Dental students (undergraduate and postgraduate) | AI virtual patient simulation | Traditional training | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Network meta-analysis vs Traditional_Simulation: Virtual_Patient_AI | SMD 0.65 [0.34, 0.95] GRADE not reported 1 study |
| Dental students (undergraduate and postgraduate) | AI virtual patient simulation | Traditional training | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Subgroup by AI modality: Virtual patient | SMD 1.12 [0.80, 1.44] GRADE not reported 1 study · 178 participants |
| Undergraduate and postgraduate dental students | Generative AI (ChatGPT) simulation | Traditional simulation (traditional training; network reference) | Overall learning outcomes (composite of psychomotor, cognitive and affective measures) Post-intervention (immediate to 1 week) · Dental education (preclinical and clinical training) · Network meta-analysis (frequentist, random effects, star-shaped network) | SMD 0.67 [0.37, 0.96] GRADE not reported 1 study · 187 participants |
| Dental students (undergraduate and postgraduate) | Generative AI (ChatGPT) guided simulation | Traditional training | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Subgroup by AI modality: Generative AI (ChatGPT) | SMD 0.67 [0.37, 0.96] GRADE not reported 1 study · 187 participants |
| Dental students (undergraduate and postgraduate) | Intelligent tutoring system (ITS) VR simulation | Traditional training | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Subgroup by AI modality: ITS | SMD 1.86 [1.07, 2.66] GRADE not reported 1 study · 36 participants |
| Dental students (undergraduate and postgraduate) | Intelligent tutoring system (ITS) VR simulation | Traditional training | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Network meta-analysis vs Traditional_Simulation: ITS_VR | SMD 1.91 [1.12, 2.69] GRADE not reported 1 study |
| Dental students (undergraduate and postgraduate) | VR haptic AI simulation | Traditional simulation (network reference) | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Network meta-analysis vs Traditional_Simulation: VR_Haptic | SMD 1.53 [0.53, 2.52] GRADE not reported 1 study |
| Dental students (undergraduate and postgraduate) | VR adaptive AI simulation | Traditional simulation (network reference) | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Network meta-analysis vs Traditional_Simulation: VR_Adaptive | SMD 1.49 [1.09, 1.88] GRADE not reported 1 study |
| Dental students (undergraduate and postgraduate) | Intelligent tutoring system (ITS) VR simulation | Generative AI (meta-regression reference category) | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Mixed-effects meta-regression by AI modality | Meta-regression coefficient (SMD difference) 1.20 GRADE not reported Studies not reported |
| Dental students (undergraduate and postgraduate) | VR-based AI simulation | Generative AI (meta-regression reference category) | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) Post-intervention (mostly immediate; one study at 1 week) · Dental education (preclinical and hybrid training) · Mixed-effects meta-regression by AI modality | Meta-regression coefficient (SMD difference) 0.82 GRADE not reported Studies not reported |
Included primary studies 18
Open in Studies| Study | Year | Also included by | Outcomes it informs here | Trial data |
|---|---|---|---|---|
| Adnan 2023 AI-enabled virtual reality systems for dental education DOI:10.53730/ijhs.v7ns1.14350 | 2023 | this review only | not stated | not extracted yet |
| Collaco 2021 Immersion and haptic feedback impacts on dental anesthesia technical skills virtual reality training DOI:10.1002/jdd.12503 | 2020 | this review only | not stated | not extracted yet |
| Collaco 2021 Immersion and haptic feedback impacts on dental anesthesia technical skills virtual reality training. PMID:33274441 | 2021 | this review only | not stated | not extracted yet |
| Huang 2025 Exploring the Application Capability of ChatGPT as an Instructor in Skills Education for Dental Medical Students: Randomized Controlled… DOI:10.2196/68538 | 2025 | this review only | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) | not extracted yet |
| Jones 2025 Artificial Intelligence Chatbots as Virtual Patients in Dental Education: A Constructivist Approach to Classroom Implementation DOI:10.1111/eje.13135 | 2025 | this review only | not stated | not extracted yet |
| Jones 2025 Artificial Intelligence Chatbots as Virtual Patients in Dental Education: A Constructivist Approach to Classroom Implementation. PMID:40460020 | 2026 | this review only | not stated | not extracted yet |
| Or 2025 Enhancing Dental Students' History‐Taking Skills With a Generative Artificial Intelligence Chatbot DOI:10.1002/jdd.13952 | 2025 | this review only | not stated | not extracted yet |
| Or 2025 Enhancing Dental Students' History-Taking Skills With a Generative Artificial Intelligence Chatbot. PMID:40500171 | 2026 | this review only | not stated | not extracted yet |
| Prinz 2025 Endodontic diagnostics training in undergraduate dental education: An observational pilot study on AI‐driven virtual patient e‐learning DOI:10.1111/iej.14277 | 2025 | this review only | not stated | not extracted yet |
| Prinz 2025 Endodontic diagnostics training in undergraduate dental education: An observational pilot study on AI-driven virtual patient e-learning. PMID:40692414 | 2026 | this review only | not stated | not extracted yet |
| Rodrigues-Pereira 2025 Generative Artificial Intelligence–Driven Clinical Case Simulation in Temporomandibular Disorder Education: ChatGPT Versus Real Patients DOI:10.1002/jdd.70104 | 2025 | this review only | not stated | not extracted yet |
| Rodrigues-Pereira 2025 Generative Artificial Intelligence-Driven Clinical Case Simulation in Temporomandibular Disorder Education: ChatGPT Versus Real Patients. PMID:41229082 | 2026 | this review only | not stated | not extracted yet |
| Sherif 2026 Adapt VR in dental education: boosting preclinical skill and self-confidence DOI:10.1038/s41405-025-00390-0 | 2026 | this review only | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) | not extracted yet |
| Suarez 2022 Using a Virtual Patient via an Artificial Intelligence Chatbot to Develop Dental Students’ Diagnostic Skills DOI:10.3390/ijerph19148735 | 2022 | this review only | not stated | not extracted yet |
| Suarez 2022 Using a Virtual Patient via an Artificial Intelligence Chatbot to Develop Dental Students' Diagnostic Skills. PMID:35886584 | 2022 | this review only | not stated | not extracted yet |
| Usta 2026 Virtual Reality Haptic Simulators and Mobile Applications—Potential AI-enhanced Tools for Improving Clinical Endodontic Training: A… DOI:10.1016/j.joen.2025.11.005 | 2026 | this review only | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) | not extracted yet |
| Vannaprathip 2025 SDMentor: A virtual reality-based intelligent tutoring system for surgical decision making in dentistry DOI:10.1016/j.artmed.2025.103092 | 2025 | this review only | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) | not extracted yet |
| Zhang 2026 The Impact of Artificial Intelligence-Enabled Virtual Simulation on Clinical Reasoning in Dental Students in Managing Temporomandibular… DOI:10.21203/rs.3.rs-8462289/v1 | 2026 | this review only | Overall learning outcomes (composite of psychomotor, cognitive and affective measures; Kirkpatrick levels 1-3) | not extracted yet |
Review question
| Element | Planned (methods) | Found (results) |
|---|---|---|
| Population | Undergraduate dental students, postgraduates or interns | Dental students, mostly undergraduate (n=12 studies; over 1,400 participants; 557 in meta-analysis) |
| Intervention | AI-driven simulation (algorithm-based decision-making, adaptive or automated feedback, simulation-based environments): VR with haptic feedback, intelligent tutoring systems, generative AI, virtual patients | AI-driven simulation: haptic VR, adaptive VR, ITS in VR, ChatGPT, virtual patient, mobile AI app (5 studies, 6 comparisons in meta-analysis) |
| Comparison | Traditional teaching (lectures, phantom-head training), non-AI simulation or standard curriculum for quantitative synthesis; no comparator allowed for narrative synthesis | Traditional training, phantom-head, lecture/practice, video-only (meta-analysis); none or real patient in other studies |
| Outcomes | Unified learning outcomes (psychomotor, cognitive, learner-reported affective) per Kirkpatrick model | Learning outcomes composite (psychomotor, cognitive, affective); pooled SMD 1.20 (95% CI 0.73 to 1.67) |
| Study designs | RCTs, quasi-experimental, controlled before-after, observational and pre-post studies | 4 RCTs and 1 quasi-experimental study in meta-analysis (n=5); 12 studies overall (4 RCTs, 2 quasi-experimental, 2 pre-post, 4 other) |
| Timing | Not specified | Immediate post-intervention to 1 week in the meta-analysis |
AMSTAR 2 items
Two independent extractors rated each item. Their ratings are shown side by side because they have not been reconciled yet. The overall rating follows Shea et al. 2017: one critical flaw gives Low, more than one gives Critically Low.
| Item | Extractor A | Extractor B | Agree |
|---|---|---|---|
| 1Research question includes PICO components | Yes | Yes | yes |
| 2Protocol established before the reviewcritical | Yes | Partial Yes | no |
| 3Choice of study designs explained | Yes | Yes | yes |
| 4Comprehensive literature searchcritical | Partial Yes | No | no |
| 5Study selection in duplicate | Yes | Yes | yes |
| 6Data extraction in duplicate | Yes | Yes | yes |
| 7List of excluded studies with reasonscritical | No | No | yes |
| 8Included studies described in detail | Yes | Yes | yes |
| 9Risk of bias assessed satisfactorilycritical | RCT: Yes · NRSI: Yes | NRSI: Yes · RCT: Yes | no |
| 10Funding of included studies reported | No | No | yes |
| 11Appropriate meta-analysis methodscritical | RCT: No · NRSI: No | NRSI: No · RCT: Yes | no |
| 12Impact of risk of bias on the pooled results | No | No | yes |
| 13Risk of bias accounted for in interpretationcritical | Yes | Yes | yes |
| 14Heterogeneity explained | Yes | Yes | yes |
| 15Publication bias investigatedcritical | Yes | Yes | yes |
| 16Conflicts of interest reported | Yes | Yes | yes |
| Overall | Critically Low | Critically Low | yes |